Re: [Inform] Inform History

[Graham Nelson <graham@gnelson.demon.co.uk>]

In article <RuLb4.9286$ok.15772@news3.mco>, Happy Poster
<URL:mailto:happy@poster.org> wrote:
> I'm trying to write an Inform manual for beginners to intermediate folks. In
> the beginning, I want to have a short word about Infocom, z-code, and how
> Inform came to be. By piecemealing things I have read, and your history on
> Graham Nelson's website, I came up with the following. Please let me know if
> anything is not factual or if you think it could be better expressed.

I thought this was a private email to me, until I found it here.
But I'm afraid the history is a little misleading, so here goes:

> "The language has an interesting history. It all started with a company
> called Infocom. Infocom was formed by some MIT students in the 70's who saw
> the work of earlier researchers in the realm of artificial intelligence and
> language parsers. Like 70's startups, they decided to make a fun company
> where dress was casual, sodas were free,

There is nothing very 70s about this, and in any case, the casual
living and late nights culture came later -- the Infocom "company
experience" belongs more or less solely to the period 1983-6.

> ... However, as computers became
> more advanced, text-based games appeared antique to many people and they
> started moving to more graphical interfaces like Apple Macintosh, Unix'
> X-Windows, and Microsoft Windows. Infocom was about to go bankrupt when a
> company out of California, Activision, decided to buy the company.

I can't agree, I'm afraid.

The buyout was June 1986, well before Windows existed.  X-Windows
hardly existed in any remotely domestic market; the Mac (c. 1984)
was very much a novelty in its user interface.  So I don't believe
this was a factor.  Infocom was not on the point of "bankruptcy",
but in any case its difficulties were caused by a venture into
business software which did not work out.  The actual games, please
note, had had an amazingly strong trading year in 1985, dominating
the SoftSel computer game distribution charts.  So there is no
reasonable argument that a collapse in popularity of the games
caused the buyout.  (Popularity did collapse, but later, and in
the opinion of some people at least, as a result of changes made
during the buyout, notably to distribution.)  Here is the relevant
section of my own history:

"Infocom's intention to explore byways of the new medium was genuine, 
but not of course altruistic, and its business history throws a good deal 
of light on its decisions. The extent of Infocom's commercial success is 
often exaggerated, not in its scale (at one time a quarter of U.S. homes 
owning computers had bought the product) but in its duration. Typical 
sales per new title rose from 10,000 in 1981 to 50,000 in 1983-6, falling 
below 20,000 again in 1988-9. The exceptions were the Zork trilogy, 
which sold 1,000,000 units over the decade - which explains if not 
excuses the later sequels - and The Hitchhiker's Guide To The 
Galaxy at 250,000, which explains Infocom's eagerness to write The 
Restaurant at the End of the Universe, a project frustrated by Douglas 
Adams's inability to get out of the bath when copy deadlines loom - 
"you can't fault him for hygiene in a crisis" (Geoffrey Perkins). Sales 
were further buttressed by customer loyalty, carefully nurtured by large 
direct mail shots (at end of 1986, circulation of the newsletter was 
240,000); by repackaging of 1980-2 titles; and by a no-returns policy in 
distribution (ended in 1987) obliging shopkeepers to treat Infocom's 
wares as luxury goods, kept on shelves until they sold. Remarkably, 
Suspended (1983), not an obvious money-spinner, was to receive a 
Gold certificate for 100,000 sales in 1986: typical shelf times today are 
measured in months or even weeks. Infocom's customers were, 
according to market research, adult (75% over 25) - which is not so 
surprising given prices of $40 to $50 - and heavy readers, 80% of them 
men, though specific products were designed to appeal to women (such 
as Plundered Hearts and the mysteries) and to children (Stu Galley 
adapted the Seastalker parser to children's sentence structures, observed 
during testing). The work force had grown fast (1981, two; 1982, four; 
1983, twenty; 1984, fifty; 1985, one hundred) but was increasingly 
preoccupied with managing itself and with Infocom's one business 
product, the database Cornerstone (1986). It was intended to capitalise 
on Infocom's expertise in virtual machines, which allowed large 
programs - adventure games - to run on a variety of different designs of 
small computer: but this was not the strength in 1986 that it would have 
been in 1982, since the IBM PC had grown in capacity and cornered the 
business market, most of the rival manufacturers having gone bust in 
1983-4. Cornerstone sold 10,000 but only after a price reduction from 
$495 to $100, and by then Infocom had turned the corner into loss. In 
June 1986 Activision had bought Infocom, in what amounted to an 
agreed merger, for stock valued at around $8 million: at about five years' 
gross income, this was a high price, or would have been if the stock had 
in fact been worth that. Infocom still had fifteen titles ahead, including a 
few of its best, but disputes over branding, marketing and the division of 
profits and losses produced disquiet, while Activision had its own 
travails. By 1988, though, market conditions would have obliged any 
management to salvage the Infocom brand-names but abandon text for 
largely graphical games. The company now called Activision (there was 
a second, happier merger) has duly done this, but a pleasing footnote is 
that its omnibus 1990s reissues of the text games achieved unexpectedly 
high sales."

> Activision is still in place today, and they have made a 3D graphical
> version of Zork.

Three, at least.

> Zork was originally a series of very popular text
> adventures revolving around a dungeon. However, let's step back a bit. In
> its height, Infocom wrote an assembler language that when compiled made a
> game file called a z-code file. They patented it.

Is that so?  I'm not aware of any patent.  I rather suspect that such
a patent ought to have fallen foul of prior art from Pascal P-code, but
of course U.S. patent courts have still only barely learned the first
thing about computing, so who knows.

Also, "in its height" is wrong.  Z-code was invented in 1979 when
Infocom had existed for about a week and employed zero people: its
sales figures were, I think, 1.  (A tape of a mainframe version of
Zork.)

> The z-code file was a very
> compressed format because hard drive space was expensive.

Hard drive space was non-existent: floppy drive space was expensive.

> You would open
> z-code files with an interpreter which then ran the game.

A detail, but they were of course seamlessly made into a single
program as far as the user was concerned.

> When Infocom
> merged with Activision, the z-code file was a thing of the past

Not so.  In 1987-9 they released many of their better works, using
Z-code, so it certainly wasn't a thing of the past.

> and they let
> the patent expire.

Again, please give references: it's news to me.

> Then, a clever computer programmer

ahem, an idle maths PhD student who ought to have been worrying
more about perturbations of the Chern-Simons equations on a
three-manifold with boundary...

> in the United Kingdom
> named Graham Nelson reverse engineered the format

No, the hard work was all done by a great many other people,
including Mark Howells, Mike Threepoint and the InfoTaskForce
group.  I really contributed very little to the decipherment.

> and using a Unix C
> language compiler building tool, awk, created a language called Inform and a
> compiler as well.

At no time have I ever used "awk" for anything.  Inform has never
used any compiler-building tool: its present version is a 1-pass
compiler optimised by hand for speed and low memory consumption
(because these were major issues circa 1995, when home computers
were about five times smaller and slower than they are today).

> He decided to make the tool free. Others started building
> the z-code interpreters

Z-code interpreters existed long before Inform did: in particular
"Zip" and "InfoTaskForce" interpreters certainly did, and they
weren't the only ones, though they were the most complete.

-- 
Graham Nelson | graham@gnelson.demon.co.uk | Oxford, United Kingdom